Talk:Icelandic phonology

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Content change[edit]

I’ve moved the phonological content from the main Icelandic language article to here, leaving copies of the tables and a brief summary there. I’ve moved the old content of this page to Icelandic orthography, where it is better suited. Max Naylor 18:16, 27 April 2007 (UTC)


Is there a phonemic contrast between the "plain" laterals [l, l̥] and the velarized laterals [l̥ˠ, lˠ]? None of my (admittedly very scanty) resources indicates that L is ever velarized in Icelandic, let alone phonemically so. —Angr 15:43, 12 June 2007 (UTC)

Well, I was also surprised to see it but it seems that for some Icelanders "veldi" ((I would) choose) and "velgdi" ((I would) make warmer) are not homophones. I believe the reference comes from the book "Íslensk tunga" [1] but I don't have it. Perhaps someone who has the book close by will come and quote the relevant passage. Stefán 19:14, 12 June 2007 (UTC)

Voicing contrast[edit]

The current text says:

Icelandic has an aspiration contrast between plosives, rather than a voicing contrast, something relatively rare among European languages

Is it really? I'm pretty sure it's the exact same in Danish. The difference between Danish /p/ and /b/ is not one of voicing, it's one of aspiration. --Pinnerup 23:49, 29 July 2007 (UTC)

Well, it says relatively rare. Danish certainly is another example, Faroese probably another. Those are the only ones I am aware of. If there are more then we should consider rewording the statement. Stefán 03:52, 30 July 2007 (UTC)
Scottish Gaelic is another one; it's actually not rare in Northwestern Europe. —Angr 04:52, 30 July 2007 (UTC)
Right, I've removed it, and another paragraph which seemed to be confusing the written and the spoken language. We still say that a voicing contrast in nasals is rare among the languages of the world. Is that an accurate statement? Also, we say that length is contrastive for consonants, but not vowels. Am I correct in saying that length is only contrastive for [m], [n], [r] and [s]? I cannot think of any others. Also this may not be terribly accurate, the research of Pind suggests that it is the relative length of the vowel compared to the following consonant rather than the absolute length of either one which is the important factor. There is also eyddra vs eitra in which either the vowel length is contrastive or my statement above about which consonants can be contrastive is wrong. Comments? Stefán 16:30, 30 July 2007 (UTC)


This page has a lot of example words with no translation. It would be a good thing to translate the example words. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 04:58, 30 August 2007 (UTC)

Agiesh hulumar Johannsson the great (talk) 14:36, 23 April 2017 (UTC)

J in the phonology[edit]

j is an approximant in the IPA. Why is it put in the fricative bit here? Also, j is included as being one of the fricatives which veer towards being an approximant here but in 'Icelandic language' it is not included. Munci (talk) 18:46, 3 May 2008 (UTC)


Are the unusual diphthongs rising or falling? That is, where should we put the <  ̯> diacritic, and how does that fit the length diacritic? kwami (talk) 20:57, 3 June 2008 (UTC)

Glottalization of non-aspirated stops?[edit]

This abstract seems to be saying that the difference between aspirated and unaspirated consonants in Icelandic involves the glottis being open (or closed) during articulation of the consonant. This would suggest that at least some unaspirated Icelandic stops are in fact glottalized or ejective. I don't speak Icelandic, but I occasionally see Icelandic shows on my cable TV (the SCOLA channel), and I can definitely hear some glottalized /k'/ sounds (which have a distinctive "click" or "pop" sound). Thoughts or comments on this from anyone who does speak Icelandic? Richwales (talk) 05:13, 18 January 2009 (UTC)

If the ejective quality were that strong and clear I'm sure it'd have been noticed already. I rather suspect that it is more like in Eastern Armenian, where the unaspirated voiceless stops have a quality that almost sounds like ejectives but not quite, and I've heard they're better described as tense. That would make them similar to the Korean tense stops, though phoneticians don't seem to be quite sure if tenseness, stiff voice or faucalized voice is the best way to describe their distinctive quality. --Florian Blaschke (talk) 15:13, 22 April 2010 (UTC)

Incorrect IPA[edit]

I have corrected a few important mistakes in the IPA of vowels, where it was previously written [øy]. This sound does not exist in Icelandic, and so I don't know who put it there but it's wrong. I think someone might have used AIPA thinking it meant [øj]. In any case, I've changed it to [øɪ], but if you like you can change it to [øi], though [øɪ] sounds better to me. Also, I changed a few [c] and [ç] to [cj] and [çj] (vekja was [veca], now it is [vecja] and hjá was [çau] but now it is [çjau]. There is definitively a difference in sound between "kja" and "ki" in Icelandic.

Anyways, I hope this helped. (Egein (talk) 14:59, 1 August 2009 (UTC))

You've left the 2nd change unfinish'd. You seem to have made the changes only in the consonant table, and even in the section of the table where it claims that eg. orthographic <kj> indeed is /c/ rather than /cj/! And then there's the paragraph under the phoneme table where it discusses this phenomenon, claiming that eg. góla vs. gjóla is a minimal pair.
Now, I'm sure there's a distinction between kja and ki - different vowels! - but is this one between /cʰja/ and /cʰi/, or /cʰa/ and /cʰi/? Try something like vs. ke: if you're right, these should be distinguishable as /cʰjɛ/ and /cʰɛ/.
Eliminating the [uiː] allophone of /u/, which you don't mention here, has a similar problem of leaving the table now presenting a non-distinction.--Trɔpʏliʊmblah 11:33, 24 August 2009 (UTC)

Voiceless laterals[edit]

I see many sorces representing these as fricativs rather than voiceless approximants (eg. John Wells here: [2]). Which is right, or is there variation? --Trɔpʏliʊmblah 11:54, 24 August 2009 (UTC)

Differing voiceless nasals[edit]

Is it really possible to distinguish [m̥ n̥ ɲ̊ ŋ̊] out of context? Wisapi (talk) 15:26, 26 January 2010 (UTC)

Do you mean to distinguish them from each other or to distinguish them from their voiced counterparts? Either way, I think the answer is yes, why wouldn't it be? +Angr 15:29, 26 January 2010 (UTC)
I meant distinguishing them from each other because in my attempts to pronounce them they only sounded like an air release through the nose. Or should air also come out through the mouth in [n̥ ɲ̊ ŋ̊]? Wisapi (talk) 19:56, 26 January 2010 (UTC)
No, like all nasal stops air should come out through the nose only. I guess the difference between them is really hard to hear if you just say them in isolation without any vowels next to them. +Angr 20:52, 26 January 2010 (UTC)
I tried with my wife (we're both native speakers) and she couldn't distinguish between my production of voiceless m and voiceless n out of context (I turned my face away from her). There are not many minimal pairs - I had to search for ten minutes before finding one - 'reimt' ("haunted") vs. 'reynt' ("tried"). Interesting question, I hadn't really thought about this before. Haukur (talk) 21:45, 26 January 2010 (UTC)
I tried a sentence pair with my wife - "Það hefur verið reimt þarna" vs. "Það hefur verið reynt þarna" - and we agreed that there was no particular problem in distinguishing them. I've thought of another minimal pair - 'lengt' ("extended") vs. 'leynt' ("secret") and I think the difference here is very slight. But that's not specific to the voiceless pairs, I think the distinction between 'lengd' ("length") and 'leynd' ("secrecy") is not made by all speakers and I'm not sure I make it myself. Haukur (talk) 22:09, 26 January 2010 (UTC)
I've tried the words "reimt" and "reynt" alone (not in a phrase) and I think I could tell them apart by the release after [m] (which sound as a very weak [b̥]). What do you think? Could saying them in a whole sentence give other clues? And what about [ɲ̊ ŋ̊]? Can you guys tell them apart (isolated or within a word)? From what I got in this article, 'lengt' and 'leynt' aren't exactly a minimal pair: 'leynt' has a <y> ([i(ː)]) which 'lengt' hasn't. Wisapi (talk) 23:17, 26 January 2010 (UTC)
You may be right about the release. The word 'lengt' is pronounced as if written 'leingt'- Icelandic orthography has a quirk about vowels before ng/nk ('banki' is pronounced as 'bánki', 'vinka' as 'vínka' etc.) I can't think of a minimal pair between [ɲ̊ ŋ̊] and I doubt one exists, but I think the same goes for the voiced variants. Haukur (talk) 00:14, 27 January 2010 (UTC)
I just wanted to add that we have a three way minimal pair since "Það hefur verið rengt þarna" (from the verb 'rengja') is a valid sentence with ŋ̊. I sometimes listen for the pronunciation of 'lygnt' and I think it is a bit off and on whether people have the ŋ̊ or whether it becomes n̥. In the latter case it sounds the same as 'lint', but I think there is an audible difference between the ŋ̊ and n̥ versions. Without any context I can believe that it is difficult to distinguish between these sounds. Stefán (talk) 11:55, 27 January 2010 (UTC)
Excellent point! Indeed, reynt-reimt-rengt is a three-way minimal pair (and of course clearly distinct from reynd-*reimd-rengd). Haukur (talk) 12:34, 27 January 2010 (UTC)
I think the term "minimal triplet" is more common than the oxymoron "three-way minimal pair". +Angr 12:44, 27 January 2010 (UTC)
Duh, yes, thanks. Haukur (talk) 13:37, 27 January 2010 (UTC)
I believe here too one can distinguish between 'lyngt' and 'lint' by the release of [ŋ̊], which sound as weak [ɡ̊] (or perhaps even a ligual egressive click) No vowel-before-/ng/ quirk here?. Should this be the case, I'd be interested in hearing of a minimal pair between [ɲ̊ n̥], for then, I imagine, the tongue would glide from [ɲ̊] to the alveolar ridge making no audible release, like [n̥]. Wisapi (talk) 16:03, 27 January 2010 (UTC)
The word is 'lygnt' (it is the nominative neuter form of the adjective 'lygn'), still the pronuciation is [lɪŋ̊t]. I could also believe that there is something at the onset of ŋ̊ which would help distinguish it from n̥ but that is speculation. As for a minimal pair between [ɲ̊ n̥] they don't exist as far as I believe. Certainly not in this environment. The highest peak in Iceland is called either Hvannadalshnjúkur og Hvannadalshnúkur but I think that the second component (hn(j)úkur) is [n̥(j)ukʏr]. Stefán (talk) 17:07, 27 January 2010 (UTC)
Do you mean that the pronunciation of 'Hvannadalshnjúkur' could be with a [ɲ̊] owing to the <j>? That's an important question because we will have to add this enviroment (<nj>) to the table then. It is also important to revise the vowels table because of the lyngt's pronunciation; I know it's not a big difference between [i] and [ɪ], but, as it stands now, it claims that the pronunciation should be [liŋ̊t]. Wisapi (talk) 00:47, 31 January 2010 (UTC)
Like I said, I think 'hnjúkur' is [n̥jukʏr], I was just wondering whether it could be [ɲ̊ukʏr] but I don't think it is. We don't need to modify the table because the word is 'lygnt'. It is *not* lyngt and the spelling does not contain the cluster ng, but rather the cluster gn. Compare 'hringt' [r̥iŋ̊t] from the verb 'hringja' (to call) with 'hrygnt' [r̥ɪŋ̊t] from the verb 'hrygna' (to spawn). Stefán (talk) 01:13, 31 January 2010 (UTC)
I understood that you were just wondering if the pronunciation of 'hnjúkur' could be something else, but as you left the question open, I was willing to draw attention to its importance, so more people may give their opinion. As to 'lygnt', I'm sincerely sorry that I bothered you again; I still hadn't noticed the different clusters. Wisapi (talk) 12:48, 31 January 2010 (UTC)

<ö> before <gi>, 'lögin' [ˈløːjɪn][edit]

Examining the vowels table, I found something that may be a typo, or, if not, deserves more explanation: 'lögin' was transliterated as [ˈløːjɪn] without the expected <ɪ> after <ø>. Wisapi (talk) 00:20, 31 January 2010 (UTC)

It is a typo. Stefán (talk) 01:14, 31 January 2010 (UTC)
Fixed, it now stands as [ˈløɪːjɪn]. Wisapi (talk) 12:52, 31 January 2010 (UTC)

Initial <h> before consonant, aspiration or devoicing?[edit]

I think it would be clarifying to add a row for that in the consonants table. It seems to be pronounced in both ways, but further down in the vowels table hræsni was only given the pronunciation [ˈr̥aistnɪ]. Is there a prefered pronunciation? By the way, what about <hv>? I've read in a grammar (Colloquial Icelandic, p.7 that some Icelanders consider the pronunciation [xv] to be better Icelandic. Is that outdated? Wisapi (talk) 01:08, 31 January 2010 (UTC)

I think it is fair to say that the consensus is that the initial clusters hr as in 'hrota', hn as in 'hneta' and hl as in 'hljóð' signify unvoiced sounds and not some cluster of an aspiration followed by a voiced sound. So, there are not two ways of pronouncing these words and the table should be modified accordingly. As for initial hv people pronounce it differently, either [kv] or [xv]. In the middle of the 20th century 3/4 of the population had the [kv] pronuciation and that fraction has been increasing rapidly. Some influencial people have claimed that [xv] is preferable. It is clear that they have lost this battle. Let me strongly recommend to anyone learning Icelandic to choose the [kv] since it is so much more popular. Stefán (talk) 01:34, 31 January 2010 (UTC)

Alternative transcriptions[edit]

Looking through an Icelandic grammar (Kennslubók í Nútíma Íslensku handa Ítölum by Riccardo Venturi (Rikarður V. Albertsson)), I came across a phonetic table and found some differences to the one we have here. I'll try to present all of them in the following subsections as coherent and faithful as I can, so you may understand as much as possible of this other concept. The author, his apparent precision notwithstanding, seems to have used some characters interchangeably as the ' (apostrophe) and ` (grave accent) to mark the words' initial stress (and some of them hadn't even that: "laut [löi:tʰ], háll [hau¿d’l’]" (non-stressed words?)). Occasionally, he also presented a symbol for a phoneme but then used another way to transliterate it in that phoneme's example word. Assuming that the phones represented by the first symbols are but the implicit, theoretical phoneme's form, and not the form that exist within a word, I'll chiefly use the latter form and present the former only together with the example word. This one I'll call prototypical phoneme while reserving 'archetypical' to the phones derived from removing the diacritics of another ([t']->[t]). Wisapi (talk) 17:19, 3 February 2010 (UTC)

Voiceless or ejective[edit]

One of the things I found out was that the sounds covered here as voiceless stops [b̥ d̥ g̥] were represented there as ejective consonants [b' d' g'] and the letters <p t k, b d g> were assigned to either them or their aspirated counterparts [pʰ tʰ kʰ] in the following fashion:


  1. [g']/[kʰ] is followed by back vowels and [g'ʲ]/[kʲ] by front vowels, cf. next section. There's no [kʰʲ] within a word, but only as a prototypical phoneme.
  2. Exception (which might be a typo): degi ['dei:ji].
That word is not an exception, the combination 'egi' is always pronunced like that. Haukur (talk) 12:04, 4 February 2010 (UTC)
Yes, the combination 'egi' wasn't mistranscribed, but the initial <d>, was. It should be [d']. Wisapi (talk) 00:57, 17 February 2010 (UTC)

  • in initial position:
  1. <b d g> were mapped to [b' d' g(ʲ)']
  2. <p t k> were mapped to [pʰ tʰ kʰ/kʲ]
  • in middle position:
  • after a vowel
  1. b -> not found
  2. d -> not found
  3. g -> never a stop
<gi> -> [jɪ]
before <t> or <s> -> [x]
otherwise -> [γ]
  1. p -> only found in 'skipti' (see below)
  2. t -> [tʰ] exception: gæta [`g’jai:t’a]
That word isn't an exception, an orthographic 't' between vowels is often aspirated in the north and usually not in the sourth. Haukur (talk) 12:04, 4 February 2010 (UTC)
  1. k -> [kʲ] / (possible [kʰ]?)
before <t> or <s> -> [x]
  • geminated
  1. bb -> [b':]
  2. dd -> [(¿)d’(:)] : Edda -> [`ed’:a] , þíddur -> [`θi¿d’ür] . The prototypical sound of <d> is [d’(:)].
  3. gg -> ejective g "[g']", palatalised before front vowels
[g’:] -> vagga -> ['vag’:a]
[c:] -> leggja -> ['leʰg’ʲ:a]
The pre-aspiration doesn't belong there, this looks like a transcription of *lekkja. Haukur (talk) 12:04, 4 February 2010 (UTC)
  1. pp -> [ʰb']
  2. tt -> [ʰd']
  3. kk -> [ʰg']
  • after a consonant -> all six letter were mapped to [b' d' g'(ʲ)]
exception: skipti [`skʲɪb’d’ɪ]
Should be [`skʲɪfd’ɪ]. Haukur (talk) 12:04, 4 February 2010 (UTC)
note the pronunciation of <i> in: herti [`her’d’i]
  • in final position (-> same rule as initial position?)
  • after a vowel
  1. b -> not found
  2. d -> not found
  3. g -> not found
  1. p -> not found
  2. t -> [tʰ]
  3. k -> [kʰ]
  • after a consonant -> presence of archetypically voiceless stops
  1. b -> not found
  2. d -> not found
  3. g -> not found
  1. p -> not found
  2. t -> [t']
  3. k -> [k']

I also think it's worth noting here the pronunciations I found for the words containing following clusters:

  • <ll> allt [al’t’], ell [eʰ d’l’], olli [`o¿d’lI]
núlli [`nu¿d’l:I] (archetypically: [(¿,h)d’l’], perhaps 'ell' should enter into this group), hóll [hou¿d’l’:], stóll [sd’ou¿d’l'], alla [`a¿d’l’:a]ull [ü¿d’l’], fýll [fi¿d’l’], höll [hö¿d’l’]
this should just be [`nul:I]; recent loanwords usually have [l:] rather than [tl]
Olli [`ol:I]
all the transcriptions here are ipsis litteris, including the one of 'ell' (space) and 'olli' (second 'l' is not ejective). These are probably typos.
  • <nn> inni [`In:I], enn [en:], brenna [`b’ren:a], einn [ei¿d’n’]
enginn [`éiηjgʲIn] (archetypically: [η]; acute accent reprisenting tone?)
  • <n>langa [`lauη'g’a], henti [`hen’d’I]
  • <fn> svefn [sveb’n’]
  • <m> [m(’)(:)] kempa [`kʲem’b’a]
  • <x> [xs] lax [la:xs] (the pronunciation [ks] was not listed. Does it exist only in loanwords?)
--Wisapi (talk) 17:19, 3 February 2010 (UTC)
No, it exists all over the place. Most people born after ca. 1950 use [ks], most people

born before that use [xs]. My wife makes fun of me for using [xs]. Haukur (talk) 12:04, 4 February 2010 (UTC)

[kʲ g’ʲ] instead of [cʰ c][edit]

Both archetypical pronunciations, amongst others, are given:

  • for <g>
  1. [gʲ] -> gef -> [gʲe:v]
  2. [c:] -> leggja -> ['lehg’ʲ:a]
but only [gʲ] is used:
gæta [`g’ʲai:t’a]
gætti [`g’ʲaiʰd’ɪ]
  • for <k>
  1. [k’ʲ] -> kæla -> [`kʲai:la]
  2. [c] -> reki -> [`re:kʲI]
but only [kʲ] is used:
kæla [`kʲai:la]
reki [`re:kʲI]
kempa [`kʲem’b’a]
poki [`pʰo:kʲI]
skipti [`skʲIb’d’I]

Further examples including the cluster <ng>/<nk>:

anga [`auηg’a]
Ingi [`iηjg’jI]
löng [löiη’k’]
enginn [`éiηjgʲIn]
ungur [`uηg’ür]
þyngð [θiηɣð]

And another surprising phonetic value for <g> that I hadn't found here:

  • [g’v] -> Guð -> [g’vüð]
--Wisapi (talk) 17:19, 3 February 2010 (UTC)
I am not sure if [v] is the correct transcription, but I definitely do recognise a pronunciation along those lines in this word ([k_wY:T]?), and I also seem to hear a different /l/ than the usual one between á and v in words like hálfur and álfur. I'd like some confirmation here. Oh, and the /a/ part of the á diphthong doesn't sound like the /a/ in a normal a to me, either. Skomakar'n (talk) 22:46, 23 July 2012 (UTC)


  • o -> [o] [wo:] kom [kho:m] [khwo:m]
  • e -> [e(:)] [ei] er [e:r], ferð [ferð]
  • é -> [je] él [je:l]
--Wisapi (talk) 17:19, 3 February 2010 (UTC)

Comments on the above[edit]

It looks like this book has some idiosyncratic transcriptions that don't follow IPA exactly. That's not unusual; most works describing the phonology of a single language have their own conventions that deviate in some way or another from IPA principles. For example, does the author ever explicitly call his "[b' d' g']" ejective? It's far more likely that he's just following a convention that the apostrophe indicates voicelessness in that book. The same goes for not indicating stress: many authors omit stress marks on monosyllabic words because it's obvious where the stress goes. Whether one uses the symbols for palatalized velars ( etc.) or the symbols for palatals (c etc.) is also a matter of convention; as long as the language you're describing doesn't contrast the two, it doesn't much matter which set of symbols you use. (Over at Scottish Gaelic phonology, we use the symbol ç to stand for a voiceless postpalatal or prevelar fricative but the symbols ʎ and ɲ to stand for a postalveolar or prepalatal lateral and nasal, even though ç, ʎ and ɲ are all in the same "Palatal" column of the IPA.) +Angr 18:05, 3 February 2010 (UTC)

Finally having some time to read a bit further in that grammar, I found out that what was meant by these apostrophes after [b d g] is that they are desonorizzate (dessounded), i.e. the vocal cords do not cease their vibrations before the onset of the articulation, but after. The author mentioned that he could could also have reprisented them by [b̥ d̥ g̊], but preferred the apostrophy. I think it would be a good idea to have this last set of symbols used here with the proper explanation of how they differ to simply voiceless consonants. Wisapi (talk) 02:47, 6 June 2010 (UTC)
However, I just read, in final positions or after certain consonantal clusters, these desonorizzate consonants will be truly voiceless, i.e. the vocal cords will stop vibrating before their onset. These are marked in this grammar by [pʼ tʼ kʼ]. Wisapi (talk) 12:39, 6 June 2010 (UTC)

Dialectical variation[edit]

I highly recommend for pronunciation samples. Haukur (talk) 12:07, 4 February 2010 (UTC)

Are the recordings freely licensed? If so, we can copy them to Commons and use them directly. +Angr 13:01, 4 February 2010 (UTC)
The site doesn't really say. I know the people behind it and I could try to brooch the subject but I would expect the usual hemming and hawing, discomfort and incomprehension. In any case, licencing would be complicated by the fact that some of the recordings are of people reading recent works of literature aloud. Some of it is just spontaneous talk which wouldn't be subject to copyright - but neighboring rights would come into play. Haukur (talk) 13:58, 4 February 2010 (UTC)
P.S. I assume you mean "dialectal variation", not "dialectical variation". +Angr 13:03, 4 February 2010 (UTC)
When I see Angr writing in <small></small> tags I know I've made a boo-boo. Haukur (talk) 13:58, 4 February 2010 (UTC)
Would you rather I point out your mistakes in great big letters? +Angr 17:39, 4 February 2010 (UTC)

problems with phonemes[edit]

Hi. I added a tag questioning the factual accuracy of the consonant section because it appears to confuse phonemes with allophones. Although I don't know Icelandic phonology very well, it appears that the following, at least, are not phonemes: velar laterals, velar nasals, palatal nasals, velar fricatives, glottal stop. It's also arguable whether either palatal stops or fricatives are phonemes (probably better analyzed as /k/, /g/ or /h/ followed by a front vowel or /j/); likewise for the voiceless resonants (predictably voiceless in some situations, sequences of /h/+resonant in other situations). The voiced dental fricative is likewise not a phoneme, as the article itself indicates.

It would be much better to clarify this situation, e.g. by enclosing the non-phonemes in brackets and the questionable phonemes in parens.

Benwing (talk) 04:05, 20 April 2010 (UTC)

Perhaps someone could quote the material from Scholten (2000). I wouldn't be surprised if there was, as the paragraph below the chart describes, some debate as to the phonemicity of certain sounds. — Ƶ§œš¹ [aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi] 06:53, 20 April 2010 (UTC)
I don't have Scholten but I have "The Germanic Languages", with a chapter on Icelandic. It has a table somewhat similar to the one here, described as "The Modern Icelandic consonant system". The important thing about the table in this book is that (a) it's never described as a table of phonemes, and (b) the book explicitly says:
The system represented here is not strictly phonemic in the classical sense nor is it the system of underlying segments in the generative sense since in contains a number of segments that are predictable (for the most part at least) in terms of their environment. This is true for the alternation between palatal and velar stops, on the one hand, and voiced and voiceless sonorants, on the other.
I wonder if the table in Scholten isn't similar, in that it's a table of major allophones, not of phonemes.
I also wonder if some of the reason for this unwillingness to commit to a phonemic representation is due to the fact that simplistic phonological theories have difficulty with a sound that functions as the allophone of two different phonemes. For example, the "voiceless r" allophone is an allophone of /r/ before /p,t,k,s/ and word-finally after voiceless consonants; but it's also the common pronunciation of the cluster /hr/.
Benwing (talk) 08:12, 20 April 2010 (UTC)

<u> in "-num"[edit]

Hi! I found in the Icelandic grammar "Kennslubók í Nútíma Íslensku handa Ítölum by Riccardo Venturi (Rikarður V. Albertsson)" that <u> is pronounced like an <o>. Unfortunately, the author uses the same character for both [o] and [ɔ], so I don't to which he's actually referring. The examples he gives are "dölum" ([ˈdʼöːlom]) and "himnunum" ([ˈhimno:nom]). Here, however, we have that this sound ([ɔ]) is only found in in the first <u> of the ending "-unum", as in "augunum" [ˈøɪːɣɔnʏm]. Wisapi (talk) 22:08, 5 June 2010 (UTC)

götva [ˈkœːtva][edit]

Shouldn't this be pronounced [ˈcœːtva], since <k> is followed by a front vowel [œ]? Wisapi (talk) 23:44, 5 June 2010 (UTC)

No, that rule only applies to unrounded front vowels. – Krun (talk) 10:35, 20 July 2010 (UTC)

hver and compounds[edit]

I've just hit upon this particularity: according to this Kennslubók, 'hver, 'einhver' and 'sérhver' "should actually be pronounced -[kʰvʏr] ([kwʏr])" (actually the author uses [ü] instead of [ʏ]). Suspecting this would be a prescripted —and not descriptive— pronunciation, I was also puzzled by the fact that the auther didn't then prescript the [xvʏr] pronunciation. Wisapi (talk) 00:37, 6 June 2010 (UTC)

aspirated consonants in non initial positions[edit]

Still according to this grammar there actually are aspirated [k p t] in middle and final positions. Examples vaka [ˈvaːkʰa], vík [viːkʰ], skapa [ˈsgʼaːpʰa], glöp [gʼlœːpʰ], gráta [ˈgʼraːtʰa] (I guess there was a typo here with the diphthong), net [neːtʰ]. There is even a whole topic about this in the grammar and before that the author had already used transcriptions with aspirated in other positions. Wisapi (talk) 02:19, 6 June 2010 (UTC)

IPA table[edit]

I've set up WP:IPA for Icelandic to go along with the {{IPA-is}} template. Needs to be reviewed, esp. the L's.

Also, I changed the transcription of long diphthongs. I assume what was meant was that the entire diphthong was lengthened, but [ai:] makes it look as though it's the off-glide that's lengthened. It's pretty standard to put the length mark on the peak of sonority. — kwami (talk) 05:27, 11 November 2010 (UTC)

Glottal stop?[edit]

The glottal stop is listed in the consonant table but is not commented in the text. I'd like to see a source on that and examples of word containing it. Icelandic being a Germanic language, my gut instinct is to be very doubtful about its phonemicity. - Dingbats (talk) 10:35, 14 December 2010 (UTC)

Vowel length[edit]

Why is there a separate section "Vowel length" now when the issue is already explained in a much more detailed way in the preceding section? The section "Vowel length" would seem redundant then. Not to mention that the detailed explanation names a source, while the "Vowel length" section is completely unsourced. --Florian Blaschke (talk) 03:08, 27 June 2011 (UTC)

This whole article is a mess. If you want to fix it, go for it. — Ƶ§œš¹ [aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi] 03:40, 27 June 2011 (UTC)

Final devoicing, anyone? (re-post from orthography talk page)[edit]

I'm very surprised to find that the description in the article not only ignores but implicitly denies the obligatory phrase-final voiceless pronunciation of the fricatives ([v],[ð],[ɣ]), even though the recorded pronunciation examples clearly display it. Árnason (2011) certainly does mention final devoicing of fricatives, and, less consistenly, of sonorants (p.237), as a "pervasive postlexical effect that takes place before a pause", using the example of dag pronounced as [ta:x].-- (talk) 02:13, 7 December 2014 (UTC)

Right, he also mentions final devoicing on p.107 and gives the examples of bauð pronounced as [pœyːθ] (cf. bjóða [pjouːða]) and gaf as [kaːf] (as opposed to gefa [cɛːva]) (p.107).-- (talk) 20:37, 14 December 2014 (UTC)